Input validation and sanitization in Agentic AI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Why is input validation important in AI systems?

Imagine you have an AI model that predicts loan approvals. Why should you validate the input data before feeding it to the model?

A To increase the size of the dataset by adding random values.

B To make the model run faster by skipping some data points.

C To ensure the data is clean and within expected ranges, preventing wrong predictions.

D To avoid training the model on too much data.

💡 Hint

Think about what happens if the model gets unexpected or wrong data.

Input validation helps catch errors or unexpected values early. This prevents the AI from making wrong predictions or crashing.

❓ Predict Output

intermediate

Output of input sanitization function

What is the output of this Python function that sanitizes input text by removing digits?

Agentic AI

def sanitize(text):
    return ''.join(c for c in text if not c.isdigit())

result = sanitize('Agent007 is ready!')
print(result)

A Agent is ready!

B Agent007 is ready!

C 007 is ready!

D Agent is ready!

💡 Hint

Look at how digits are filtered out from the string.

The function removes all characters that are digits. So '007' is removed from 'Agent007'.

❓ Model Choice

advanced

Choosing a model robust to noisy input data

You have input data that may contain some errors or noise. Which model type is best to handle this situation?

A A simple linear regression model without regularization

B A deep neural network with dropout and batch normalization

C A decision tree with no pruning

D A nearest neighbor model using raw input features

💡 Hint

Think about models that include techniques to reduce overfitting and handle noise.

Deep neural networks with dropout and batch normalization help the model generalize better and reduce sensitivity to noisy inputs.

❓ Metrics

advanced

Evaluating input sanitization impact on model accuracy

You trained two models: one on raw input data and one on sanitized input data. The sanitized model has 5% higher accuracy. What does this suggest?

A Sanitizing input data improved data quality, helping the model learn better.

B Accuracy is not affected by input data quality.

C The model with raw data is always better regardless of accuracy.

D Sanitizing input data removed useful information, hurting the model.

💡 Hint

Think about how cleaning data affects learning.

Removing noise or invalid data through sanitization helps the model focus on meaningful patterns, improving accuracy.

🔧 Debug

expert

Identifying input validation bug in AI pipeline

Consider this code snippet that validates numeric input features before prediction. What error will it raise?

Agentic AI

def validate_features(features):
    for key, value in features.items():
        if value < 0:
            raise ValueError(f"Negative value for {key}")

input_data = {'age': 25, 'income': -5000, 'score': 70}
validate_features(input_data)

A TypeError: unsupported operand type(s) for <: 'str' and 'int'

B No error, function runs successfully

C KeyError: 'income'

D ValueError: Negative value for income

💡 Hint

Check which input value is negative and what the function does.

The function raises a ValueError when it finds a negative value. 'income' is -5000, so the error message mentions 'income'.